Models

A continual learning language model that trains from scratch on a single 8 GB VRAM GPU, assembling its own architecture and adapting its size to the hardware and data available rather than shipping as a fixed pretrained model.

• Models

An open-source contrastive-embedding pipeline that maps technical CVE vulnerability descriptions to formal cybersecurity compliance controls in NIST SP 800-53 and CMMC, published with its model generation, fine-tuning and evaluation code.

• Models
• Security

A small vision-language model, fine-tuned with LoRA on a 500-million-parameter backbone, that answers typed multiple-choice, ordinal-score and yes/no questions about an image with calibrated probabilities, in about 400 milliseconds on a laptop.

• Models
• Inference

A full-workflow tutorial and toolchain for training vision, hearing and smell AI algorithms and deploying them, as packaged algorithm modules, onto a specific edge AI platform.

• Models
• Tutorials

Measured benchmarks, build recipes and more than a hundred ready-to-run GGUF models for running large language models locally on AMD's Ryzen AI Max+ 395 (Strix Halo) hardware with 4-bit ROCmFP4 quantization.

• Models
• Inference

A large language model fine-tuned specifically for game development, acting as a technical director and lead architect that breaks down game system requirements and produces an executable development plan.

• Models

A small, educational Python package for fine-tuning Vision Transformer models with PyTorch, reaching 97.65% accuracy on CIFAR-10 with a simpler API than general model-zoo libraries.

• Models

A locally deployable, open-source multimodal mixture-of-experts language model focused on financial applications, compressed to under 16 GB through a custom quantization method while running on consumer GPUs.

• Models

An unofficial implementation for fine-tuning VGGT-Omega, a feed-forward 3D reconstruction model, on your own data, plus a standalone GPU-free 3D reconstruction viewer.

• Models

A diagnostic and audit library for Mixture-of-Experts models in HuggingFace Transformers that instruments router modules to catch expert collapse, routing entropy collapse and load imbalance during training.

• Models
• Testing

The companion code repository for a Manning book on adapting open-weights language models, covering prompting and RAG, LoRA and QLoRA, full supervised fine-tuning, distillation and DPO/RLHF alignment on a single GPU.

• Models
• Books

A modular PyTorch library for training and deploying Bangla language models from scratch, built around modern architecture components such as grouped-query attention, sliding-window attention, rotary embeddings and sparse mixture-of-experts.

• Models

Pre-compiled, encrypted streaming speech-recognition model weights in VoxRT's .vxrt format, for consumption by the VoxRT on-device runtime on Android and iOS.

• Models
• Speech

A PyTorch Tree2Tree model that translates natural-language descriptions into source code by decoding a syntax tree of grammar rules instead of generating tokens one at a time.

• Models

A PyTorch framework for fine-tuning the Suiren family of pre-trained molecular encoder models on custom regression or classification property-prediction tasks.

• Models

A set of optimizations for Unsloth on NVIDIA DGX Spark hardware that speeds up LoRA and full fine-tuning of Qwen3.5 models by roughly 7 to 8 times, with a loss curve verified bit-identical to the unoptimized baseline within BF16 precision.

• Models

A deep learning framework that classifies 16S rRNA gene sequences into all seven taxonomic ranks in a single pass, combining a convolutional network with a BERT-style transformer and reaching 92.95% species-level accuracy on a public benchmark.

• Models

A QLoRA fine-tune of Qwen2.5-VL-3B that reads a driving-scene photo and outputs structured analysis of lane lines, vehicles, traffic signs and driving risk.

• Models

A small transformer variant that merges the attention layer with the SwiGLU feed-forward layer to cut parameters at very small model sizes, without losing much performance.

• Models
• Papers

Fine-tunes IndoBART or mT5 to summarize Indonesian news articles into a headline and a short paragraph, trained on the IndoNLG and Liputan6 datasets.

• Models

An early-stage toolkit for generating synthetic datasets for LLM fine-tuning, with multiple backends, filtering and GPU acceleration.

• Models

The official implementation of a CVPR 2025 paper on retrieval-augmented neural fields for embedding geographic coordinates, with a pip-installable package that returns high-resolution visual-feature embeddings for any latitude and longitude.

• Models
• Papers

A mixture-of-experts model compression toolkit that prunes, merges and quantizes large MoE language models into smaller specialist models for local inference, implementing REAP expert pruning with several scoring strategies and a self-contained evaluation harness.

• Models
• Inference

A near-lossless 5-bit transformer compression method with verifiable, reproducible reconstruction: a CLI generates text against a hosted compressed model, browses a model catalog, and verifies a downloaded pack's integrity against a published SHA-256 manifest.

• Models
• Inference